Reflooding with internal boiling of a heating model porous medium

advertisement
Open Archive TOULOUSE Archive Ouverte (OATAO)
OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.
This is an author-deposited version published in: http://oatao.univ-toulouse.fr/
Eprints ID : 15884
To link to this article : DOI:10.1016/j.ijheatmasstransfer.2016.04.013
URL : http://dx.doi.org/10.1016/j.ijheatmasstransfer.2016.04.013
To cite this version :
Sapin, Paul and Gourbil, Ange and Duru, Paul and Fichot, Florian
and Prat, Marc and Quintard, Michel Reflooding with internal
boiling of a heating model porous medium with mm-scale pores.
(2016) International Journal of Heat and Mass Transfer, vol. 99. pp.
512-520. ISSN 0017-9310
Any correspondence concerning this service should be sent to the repository
administrator: staff-oatao@listes-diff.inp-toulouse.fr
Reflooding with internal boiling of a heating model porous medium
with mm-scale pores
P. Sapin a,b, A. Gourbil a,b, P. Duru a,⇑, F. Fichot b, M. Prat a, M. Quintard a
a
b
Institut de Mécanique des Fluides de Toulouse (IMFT) – Université de Toulouse, CNRS-INPT-UPS, Toulouse, France
Institut de Radioprotection et de Sûreté Nucléaire (IRSN), Division of Major Accidents Prevention, BP3 13115 St Paul lez Durance, France
a b s t r a c t
Keywords:
Nuclear safety
Hot debris bed
Boiling
Porous media
This paper presents a pore-scale experimental study of the reflooding of a two-dimensional model porous
medium. The objective is to better understand the reflooding mechanisms in play in the context of
nuclear reactor safety. The hot debris bed that forms in a nuclear reactor following a loss of coolant accident is comparable to a heat-generating porous medium. Its cooling by water reflooding involves intense
boiling mechanisms that must be modeled properly to assess mitigation procedures.
The experimental study presented in this paper focuses on the phenomenology of reflooding of a model
porous medium composed of a bank of mm-scale heating cylinders placed between two ceramic plates. A
FluorinertTM liquid, HFE-7000, is injected at a temperature close to saturation into the dry and superheated porous medium. Each cylinder of the test section is used both as a heating element and a temperature probe, which enables to track the evolution of the three different macroscopic zones identified
during cooling of the system. The reflooding dynamics, in particular the cooling fronts velocities, are thus
determined thanks to pore-scale thermal measurements together with direct visualizations. The influence of the injection flow rate and of the heating power are studied in a parametric way.
1. Introduction
In the aftermath of a loss of coolant accident (LOCA scenario),
residual decay heat in a nuclear reactor core can cause dry-out of
water in the vessel, destruction of fuel rods and melting of core
materials, thus creating a molten pool commonly called corium.
To prevent the spread of molten materials, the reactor vessel can
be cooled down by reflooding with water. Yet, quenching of very
hot rods in the upper plenum may result in their collapse, thus
forming a debris bed heated up by residual decay heat, comparable
to a highly permeable porous medium: the typical particle size, dp ,
is in the range 1–10 mm, which is of the same order of magnitude
as the water capillary length. Reflooding of this dry and heatgenerating porous medium is a crucial step in the cooling process.
Indeed, a non-coolable particle bed might melt down, thus expanding the corium pool and possibly damaging the reactor vessel.
In terms of nuclear safety analysis, intense boiling mechanisms
and complex flow regimes encountered at the pore scale in the
debris bed must be modeled properly to estimate the efficiency
of the reflooding and to assess mitigation procedures in case of
such severe accident. Many experimental studies have been undertaken in the past decades for understanding boiling in porous
media under various conditions: pool boiling [1–5], coolant entry
from the top [6–9] or bottom [6,10–15] to cite a few. Of particular
interest in such experiments is the determination of dryout or
reflooding conditions and model validation. A classical modeling
approach consists in describing this complex three-phase system
(two-phase cross-flow with liquid–vapor phase change within a
heated packed particle bed) at a macroscopic scale by writing
macro-scale conservation equations for an equivalent continuous
media, whose properties at each point are effective transport properties representative of the pore-scale configuration. Given the
thermal constraints observed in debris bed cooling problems, several major modeling issues must be addressed. What is the structure of macro-scale momentum equation and energy balance?
May the models be determined in an uncoupled manner, i.e., is it
possible to determine two-phase flow equations and effective
properties independently from the pore-scale boiling mechanisms? Most models make use of such an assumption. Twophase flow equations are based on generalized Darcy’s laws [16],
Ergun’s extensions taking into account inertia effects [17,4,9],
equations with interfacial drag terms [18,19]. Interestingly, the
model in [19] introduces the idea of different pore-scale organization of the phases depending on the flow rates and saturation,
Nomenclature
dc
dp
e
Gl
Lhtl
Q
S
tc
tqf
cylinder diameter (RTD probe diameter), m
diameter of the particles formed by the collapse of fuel
rods, m
porosity, –
flow rate, m3 =s
vertical extent of the heat transfer layer, m
heating power per cylinder, W
setup cross-section area, m2
time when a cylinder is said to be reflooded or cooled
down, s
time of arrival of the quenching front, s
repartition possibly dependent upon the pore-scale boiling mechanisms. However, all these models rely up to a certain point on
some heuristic views of the pore-scale physics, as well as on some
heuristic upscaling rules that deserve further investigations. The
heat transfer macro-scale equations in most models are based on
a macro-scale energy balance written assuming thermal equilibrium [16,20], while based on some experimental evidence, nonequilibrium models were also proposed [21,22,14].
Indeed, thermal inertia of the particles, important convective
heat transfers and rapid evaporation processes are responsible
for significant local differences between the three phaseaveraged temperatures defined for the solid, liquid and gas phases.
Such a local thermal non-equilibrium situation requires the introduction of separate energy equations for the different phases.
Duval et al. [23] derived macroscopic equations from the local
energy and mass conservation laws with the volume averaging
method [24]. They obtained macroscopic laws for the gas, liquid
and solid phases, accounting for diffusion, convection and evaporation contributions. Providing explicit relations between the effective properties of the macroscopic model and the pore-scale
configuration requires to write local closure problems to be solved
over representative unit cells.
So far, the theoretical results involve several assumptions that
are not necessarily compatible with the various pore-scale flow
mechanisms. The major assumptions which require further investigations are listed below:
Pore-scale two-phase flow configurations are assumed to be
known a priori in the heat transfer upscaling problem. Furthermore, the interface is assumed to obey a quasi-static assumption [25].
The time-scales are separated enough so that the time convolutions that arise in the upscaling process can be discarded
[26,27]. The resulting model is a three-temperature model.
Obviously, these assumptions do not seem compatible with
intense boiling, characterized by rapid movements of the phases
and interfaces. Finally, because of the lack of pore-scale knowledge
of the various configurations, the effective properties have been
determined analytically for very simple unit cells, considering
two main configurations: stratified solid–liquid–gas (SLG, where
liquid is the wetting phase) or solid–gas–liquid (SGL, where gas
is the wetting phase) flows, this latter configuration being reminiscent of a film boiling situation. These simple calculations emphasized the strong impact of the phases repartition upon the
effective properties.
Practical application of these developments to convective boiling problems in porous media faces a major difficulty. Boiling phenomena lead to various, often highly transient, liquid–gas
repartitions, especially at the solid–fluids interface where both liq-
T inj
T sat
vl
v qf
vc
yth
qf
yqf
yth
c
inlet temperature, °C
saturation temperature, °C
intrinsic liquid velocity, m/s
velocity of the quenching front, m/s
velocity of the cooling front, m/s
position of the quenching front in the y-direction deduced from thermal measurements, m
position of the quenching front in the y-direction, m
position of the cooling front in the y-direction deduced
from thermal measurements, m
uid and gas can wet the solid surface. Heterogeneous nucleation
and coalescence of bubbles are indeed complex processes that
can hardly be described by stratified configurations with static
interfaces, such as SLG or SGL configurations. While the above
analysis suggests that the complete up-scaling problem, i.e., the
coupled mass, momentum and energy balance equations, should
be reviewed, we focus in this paper on the heat exchange between
the different phases.
Comprehensive approaches have already been proposed to predict heat exchanges in boiling configurations and successfully compared to experimental data obtained on micro-fabricated surfaces.
For instance, studies by Liter and Kaviany [28] and Dhir [29,30]
allow to be optimistic that comprehensive models should be able
to estimate boiling heat transfer in complex configurations. However, to our knowledge, no correlation is currently available to estimate the convective boiling contribution to heat exchanges during
reflooding of a porous medium. Bachrata et al. [14] thus proposed a
semi-heuristic approach that consists of adding heat source terms
into the solid energy equation of Duval et al.’s model [23] to
account for nucleate, transition and film boiling contributions,
with empirical or mechanism-based correlations. The key challenge is to select adapted correlations, because of the numerous
influential parameters. In particular, Kew and Cornwell [31],
Thome [32] and Cheng and Wu [33] outlined the strong effect of
confinement on boiling phenomenology and on the resulting heat
transfer coefficient. Also, with any of the transition criteria proposed in these studies, the different characteristic flow paths in
the concerned packed debris beds can be classified as either
macroscale, mesoscale and microscale channels with respect to
two-phase processes (considering vapor and liquid water).
It is thus essential to obtain pore-scale visualization about
phase distribution that will help either:
to estimate effective transport properties in more realistic representative unit cells, and/or
to choose more carefully convective boiling correlations for
heuristic or semi-heuristic approaches.
Such information was not obtained in previous experimental
studies of reflooding of superheated porous structures, a subject
investigated experimentally since the 1980s [34,35,10,36]. In fact,
such studies are typically performed on 3D packing of superheated
beads, preventing any visualization at pore scale. They aim primarily at obtaining data at macroscopic scale (pressure, global vapor
production,. . .) that can be used to adjust parameters of some
macroscopic models (see, e.g., the insights obtained from Repetto
et al. [37], Bachrata et al. [14] and Fichot et al.’s [38] experimental
studies on Duval et al.’s modeling [23]).
The present experimental work aims at obtaining direct visualizations of the reflooding of a model porous medium, together with
thermal measurements at pore scale. The experimental setup and
techniques are first described in Section 2. The salient features of
the reflooding in such a system are then presented. First, in Section 3.1, a description of the transient two-phase flow taking place
in the porous medium during the reflooding is given. The heat
transfer mechanisms between the heated solid matrix and the
reflooding fluid are identified. Then, we turn to quantitative measurement of the propagation of the quenching and cooling front
observed during reflooding, as a function of the heated matrix temperature and fluid injection velocity (Sections 3.2 and 3.3). Conclusions and perspectives for further work are then drawn in
Section 4.
2. Experimental setup and measurement techniques
2.1. Experimental setup
The experimental test section consists of a nearly twodimensional porous medium made of 392 cylinders placed
between two parallel ceramic plates and held vertically, see
Fig. 1. The spacing between the two plates is 3 mm. The 2.1 mm
diameter cylinders are positioned in a slightly randomlydisturbed staggered arrangement that spans the setup width
(50 mm) and extends over a 80 mm height. The resulting porosity
is then e ¼ 0:65. The two plates are respectively a MacorTM ceramic
plate and a polished, transparent VycorÒ glass plate which seals
the test section while allowing visualization. These materials have
been chosen because of their ability to withstand high temperatures. The porous section is preceded by a divergent part to obtain
a flat liquid injection front. The two-phase mixture exits at the top
of the test section.
The key feature of the experimental test section lies in the fact
that each cylinder is a Pt-100 resistance temperature detector
(RTD) composed of a coiled platinum wire wrapped around an
insulating core, sealed into a 2.1 mm diameter and 7 mm long
ceramic coating (see Fig. 1). Each RTD is used as a heating element
Fluids outlet
(liquid and/or vapour)
but also as a local temperature probe, as explained in Section 2.2.1.
Note that the smaller size of the heating cylinders and their strong
confinement between the two ceramic plates are two major differences with typical tube bundle geometries, for which convective
boiling has been studied experimentally since the early 1950s
(see, e.g., [39–41]).
HFE-7000 was chosen as the working fluid because of its low
saturation temperature (T sat ¼ 34 C at atmospheric pressure).
Reflooding experiments can then be run with moderate thermal
constraints. In addition, this FluorinertTM fluid is non-toxic, nonflammable and dielectric, which guarantees safety during handling. HFE-7000 perfectly wets the ceramic surfaces it is in contact
with (contact angle close to 0°). Also, it is worth mentioning that
the ratio of the capillary length of HFE-7000 to the diameter of
the cylindrical heating elements is similar to the one obtained
when considering the ratio between the capillary length of water
at 100 °C and a characteristic 5 mm particle size, typical of the
one observed in debris bed. Indeed, observations of the damaged
TMI-2 reactor [42] and different experimental studies (see, e.g.,
the TMI-2 core sample examinations by McCardell et al. [43], the
LOFT experiment analysis by Coryell et al. [44], and the microscopic observations of fuel samples made by Manzel and Walker
[45]) characterized the particles formed by the collapse of fuel
rods: their size dp approximately lies between 1 and 10 mm. Thus,
realistic boiling conditions are obtained in the present experimental setup, as far as bubble size and confinement effects are
concerned.
The fluid flow is driven by a gear pump (see Fig. 2(a)), the head
of which is composed of stainless steel with PTFE teeth. The pump
thus requires no lubrication and this cancels the risk of HFE-7000
contamination by a lubricating fluid. The fluid temperature is controlled by circulation through a coil immersed in a constant temperature bath. Inlet temperature T inj and flow rate Gl are
measured upstream of the test section. For the experiments
reported in the present paper, the fluid was injected at a temperature close to its saturation temperature (T inj T sat ). The intrinsic
injection velocity v l is obtained by dividing the imposed flow rate
by the product of the setup cross-section S (3 50 mm2 ) times the
porosity e:
vl ¼
Gl
eS
:
ð1Þ
In the present study, using v l rather than the flow rate Gl is
more relevant as v l is the theoretical average rising velocity of
the liquid front in the test section, in the absence of phase change.
g
2.2. Experimental techniques
y
z
x
Heating cylinder
(Resistance Temperature Detector)
Ceramic coating
Platinum wire
Liquid injection
at Tinj and Gl
Fig. 1. Sketch of the experimental test section: a ‘‘two-dimensional” model porous
medium. Each cylinder is a platinum resistance temperature detector.
2.2.1. Thermal control, temperature measurements and direct
visualizations
The electrical resistance of each RTD (100X at 0 °C) varies with
temperature according to a well-known law. RTDs are classically
used with very low currents to measure resistance, and thus temperature, while avoiding self-heating. Whereas Joule effect is usually considered as a source of error, RTDs are used here both as
heating elements and as local temperature probes. However, the
dependence of resistance upon temperature leads to an additional
challenge. In order to control the heating power transmitted to
each cylinder, the voltage applied at the ends of each resistant wire
must be adjusted continuously. All heating cylinders are connected
to a stabilized power supply through a control module comprising
a proportional-integral controller equipped with a 1000 Hz microprocessor. Each RTD therefore dissipates a well-controlled heat
flux while its temperature is recorded at a very high acquisition
rate.
(a)
(b)
Fig. 2. (a) Experimental setup: an open fluid circuit. (b) Side view of the test section, showing the visualization configuration.
Insights on the liquid and vapor phases distribution during a
reflooding experiment are obtained by direct visualization performed using a PCO Dimax high-speed camera. The frame acquisition rate is typically around 250 frames per second, with a low
exposure time (1 ms) to avoid motion blur. A white LEDs ring providing a polarized light is placed in front of the transparent ceramic
plate and aligned with the CCD (charge-coupled device) camera
optical axis, see Fig. 2(b). This optical configuration limits the shadowing effect due to the many close cylinders that would have been
obtained with a lighting from the side. Also, the amount of light
that is simply reflected at the surface of the transparent ceramic
plate is removed by a polarizing filter. This layout enables to
achieve a nearly homogeneous lighting of the test section.
2.2.2. Protocol for a reflooding experiment
Reflooding experiments are performed on an uniformly heated
model porous medium. The constant heating power applied to
each cylinder, Q, and the liquid injection velocity, v l , are the two
experimental control parameters. The protocol for a reflooding
experiment is the following.
First, the HFE-7000 tank is closed and a vacuum pump is connected at the top of the dry test section, to extract air from the
whole fluid circuit. HFE-7000 tank is then opened so that the full
system fills gently with liquid. This procedure aims at limiting
the amount of air remaining trapped in the system. Second, a heating power is applied to the cylinders to dry the model porous medium while the test section top is opened at atmospheric pressure to
avoid any pressure build-up. Finally, liquid is injected with a velocity v l in the dry porous medium saturated with HFE-7000 vapor
and heated with a constant thermal power.
It is important to note that only a fraction of the bundle of cylinders is heated during a reflooding experiment, typically 14 rows of
cylinders (203 cylinders) in the middle of the test section, see Fig. 3
(a). While limiting the total electric power to be delivered to the
electronic control units, this is enough as measurements are then
performed in a region of the system where the influence of entry
and exit effects can be neglected (that is, respectively, establishment of the flow field in the porous bed, and flow recirculation
in the top converging part of the test section, the influence of
which extends over the very last rows of the test section).
A typical initial steady-state temperature profile is shown in
Fig. 3(b) for a heating power Q ¼ 80 mW per element. The
observed thermal gradients are due to heat losses towards the side
plates (ceramic) around the cylinder bundle, to heat conduction in
those plates and to natural convection with air outside of the test
section.
3. Experimental results
3.1. Two-phase flow configuration and temperature distribution
We first describe qualitatively the two-phase flow configuration
typically observed in a reflooding experiment. Fig. 4 shows an
image taken once the liquid has reached the heated porous region.
In various experiments of quenching of superheated debris beds
with bottom injection of a liquid coolant [6,10–15], three distinct
regions were detected from temperature measurements at different locations within the bed. The present experiment provides
the first direct observation of these zones, that are from top to
bottom:
a superheated vapor-dominated zone with some isolated liquid
droplets ejected from the so-called heat transfer layer;
a heat transfer layer (as designated by Tutu et al. [10]), that is an
intense boiling zone where coalescence of bubbles is very frequent so that one can hardly discern any flow pattern;
a reflooded – or quenched – region where heating cylinders are
cooled down to near-saturation temperature, so that nucleate
boiling takes place (isolated vapor bubbles can be clearly
visualized).
A fourth zone corresponding to a one-phase liquid flow could be
observed in such experiments if the subcooling of the fluid was
higher. This condition was not fulfilled for this experiment because
the liquid enters almost at saturation temperature. Two important
findings from these observations are the presence of isolated liquid
droplets in the vapor-dominated zone and the formation of vapor
slugs in the heat transfer layer. The ejection of liquid drops at
the quenching front, also called sputtering, favors precursory cooling of the superheated solid elements ahead of the quenching
front. This phenomenon is not accounted for in the models presented in the Introduction of this paper. The formation of vapor
slugs in the heat transfer layer is reminiscent of boiling in
microchannels, for which the boiling phenomenology is different
in comparison with macrochannels.
(a)
(b)
Fig. 3. (a) Heating configuration of the model porous medium used in the reflooding experiments. (b) Initial steady-state temperature profile of the heated part of the dry
porous medium (Q ¼ 80 mW).
Liquid droplet
Quenching front
Single-phase vapour reg
+ dipersed liquid drople
Heat transfer layer
(or intense boiling zone
Reflooded region
(nucleate boiling regime
Fig. 4. Visualization of the two-phase flow configuration during reflooding. In this experiment, the porous medium was heated with a constant heating power Q ¼ 80 mW
and the liquid injected at a velocity v l ¼ 3:6 mm=s.
Superheated Vapor Region
with some liquid droplets
g
Quenching front
Heat Transfer Layer
Intense boiling zone
(nucleation, coalescence,
film boiling...)
y
yqf
x
Cooling front
Quenched Region
Nucleate boiling regime
yc
Liquid injection
at Tinj and Gl
Fig. 5. Sketch of the three distinct regions observed during reflooding, with the
definitions of the quenching and cooling fronts.
Such a two-phase flow configuration, sketched in Fig. 5, leads us
to define a quenching front separating the two first zones (it is
highlighted as a clear line in Fig. 4 cooling front separating the
two last ones, that is difficult to detect precisely from visualizations. Instantaneous temperature distribution at the test section
scale confirms the relevance of such a description, see Fig. 6. It is
noticeable that despite the high permeability of the porous medium, the thermal gradients in the initial temperature profile
(Fig. 3)) and the higher permeability at the lateral sides of the
cylinder bundle, macroscopic behavior during a reflooding experiment turns out to be very close to a one-dimensional configuration
with flat quenching and cooling fronts and very low thermal gradients in the x-direction. Consequently, quantitative information
about the two fronts displacements can be obtained by focusing
on the temperature evolution recorded for each heating element.
Fig. 7 displays a typical temperature vs time signal obtained for
a single central heating cylinder. Once reflooding starts, a significant temperature decrease is observed. The local thermal nonequilibrium between the vapor phase and the superheated solid
is not the only reason for the solid element temperature decrease
downstream of the quenching front. This precursory cooling is
Superheated vapor-dominated zone
Heating cylinders
temperature (°C)
85
Quenching front
90
80
Heat transfer layer
80
Temperature
scale (°C)
70
75
70
Cooling front
60
65
Reflooded region
50
40
60
55
50
30
40
80
70
60
30
50
40
y (mm)
20
30
20
10
10
0
0
x (mm)
50
45
40
35
Fig. 6. Instantaneous temperature spatial distribution during reflooding, at a time corresponding to the two-phase flow distribution shown in Fig. 4.
Heating element temperature (°C)
INITIAL EQUILIBRIUM
Precursory cooling
Quench front arrival
Intense convective boiling
Element reflooded
FINAL EQUILIBRIUM
Time (s)
Fig. 7. Temperature evolution of a heating element during reflooding.
partly due to the forced convection associated with the vapor flow
generated in the zone upstream from the quenching front, but also
to heat conduction through the ceramic walls and, based on the
visualizations, to sputtering of liquid drops from the quenching
front. At a time noted tqf (t qf ¼ 24:8 s in Fig. 7), the heat flux
extracted from the heating element increases abruptly. Eventually,
the cylinder is considered to be reflooded – or cooled down – at t c ,
when a permanent nucleate boiling regime is reached around it
(note that a reflooded cylinder still produces vapor because the
heat flux supplied to it is maintained constant during the experiment at a sufficiently high level to cause boiling).
3.2. Quenching and cooling fronts detection
Relating the arrival of the quenching front at the y-location of
the considered heating element to the temperature drop observed
at tqf is a priori not straightforward. Heat diffusion through the
walls and thermal inertia of the cylinders might induce a delay
between the actual position of the two-phase front yqf and the
position obtained from thermal measurements yth
qf (deduced from
the time tqf associated with the discrete spatial y-location of each
cylinder). Indeed, the response time of a RTD probe is approximately 0.4 s, which is not negligible with respect to the cooling
duration of a heating element (t qf t c ¼ 4 s in this example).
The position of the quenching front yqf can be obtained unambiguously by processing the image sequence given by the high-
speed camera. After detection of the cylinders, a mask is created
in order to isolate the pore space. A background is then subtracted
from each image of the sequence and a canny filter is used to
enhance edges (i.e. liquid–vapor interfaces in the present case).
Binary images are finally generated, from which an average
quenching front y-location can be obtained, see Fig. 8(a). In Fig. 8
(b), the temporal evolution of yqf obtained through image processing (gray dots) is compared with that of the position yth
qf derived
from thermal measurements (filled black squares). The match
between these two plots shows that the break in slope observed
at t qf in Fig. 7 truly corresponds to the arrival of the quenching
front.
With regard to the cooling front, its location is obtained from
thermal measurements only, yth
c being deduced from the time t c
associated with the discrete spatial y-location of each cylinder.
Finally, temperature measurements can be used with confidence to track the spatial evolution of the quenching and cooling
fronts locations, by determining the characteristic times tqf and tc
for each of the heated cylinders.
3.3. Parametric study of the quenching and cooling fronts velocities
As discussed above, tracking the locations of both the cooling
and quenching fronts is made possible by the analysis of the temperature evolution of heating elements. As shown in Fig. 8(b), yqf
increases linearly with time as soon as the quenching front has
travelled over the vertical extent of a couple of rows of heating elements. A similar observation can be made for the evolution of the
cooling front location yc . Therefore, obtaining the constant quenching and cooling front velocities v qf and v c is straightforward and
we now discuss their evolution with the experimental control
parameters.
First, these two fronts velocities are plotted in Fig. 9(a) as a
function of the heating power Q applied to each heating element,
for a constant liquid injection velocity v l ¼ 5:47 mm=s. Increasing
the heating power Q results in larger amount of energy to be
extracted by the two-phase quenching zone, which in turn induces
a decrease in the fronts velocities, as expected. The very important
decrease observed in Fig. 9(a) suggests that it may be possible to
reach a stationary regime for which the fronts would remain stationary. However, for the experiment performed with
Q ¼ 160 mW, the initial temperatures in the cylinders reach
200 C. Heating up further would threaten the integrity of the
epoxy resins used to seal the RTD probes into the MacorTM plate
(see Fig. 2(b)). Therefore, it was not possible to study higher power
levels.
(a)
(b)
Fig. 8. (a) Binary image used to detect the quenching front location, obtained after processing of the image shown in Fig. 4. For each column of the image, the location of the
uppermost black pixel is automatically detected. Then, an average is performed along the full width of the image to obtain yqf (shown as a yellow solid line). (b) Temporal
evolution of the quenching front location obtained from temperature measurements (filled black squares) and image processing (gray dots). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)
(a)
(b)
Fig. 9. (a) Influence of the heating power Q on the fronts velocities (v l ¼ 5:47 mm=s constant). (b) Influence of liquid injection velocity
(Q ¼ 80 mW).
Second, the fronts velocities are plotted as a function of the liquid injection velocity v l for a fixed heating power Q ¼ 80 mW, see
Fig. 9(b). As expected, increasing the liquid inlet velocity causes the
fronts velocities to increase as well. For injection velocity inferior
to 5 mm/s, the fronts velocities are similar. However, for larger
v l , the quenching front velocity is larger and the higher v l , the
more important the difference between v qf and v c is. These trends
are similar to those observed on 3D column packed beds such as
the PRELUDE facility which consists in a 174 mm internal diameter
cylinder filled with 200 mm high packed bed of steel balls heated
up with induction, reflooded with bottom injection of water at
ambient temperature [37].
Fig. 10 shows the evolution of the heat transfer layer vertical
extent Lhtl during a reflooding experiment, for the different injection velocities investigated. Lhtl is measured thanks to the thermal
measurements, as explained in Section 3.2, and is defined as
th
Lhtl ¼ yth
qf yc . On each single curve, three distinct periods can be
vl
on the fronts velocities
observed. The first period, during which the size of the heat transfer layer rapidly increases, corresponds to the ingress of the
quenching front into the superheated test section. Once the first
bottom elements of the heated part of the test section are cooled
down and have reached a steady-state nucleate boiling regime, a
cooling front propagates as well within the bank of cylinders,
which induces the first change in slope observed on each curve.
Then, both a cooling and a quenching front can be detected in
the heated part of the test section, as explained above. For small
liquid injection velocities (v l ¼ 1:25 and 3.36 mm/s), the extent
of the heat transfer layer does not vary much with time as v qf
and v c are very close to each other (see Fig. 9(b)), while it significantly increases as the two fronts propagates into the porous medium, for more important flow rates (for which v qf > v c ). Finally, the
size of the heat transfer layer diminishes once the quenching front
has reached the upper limit of the heated part of the test section,
until the cooling front reaches this limit as well.
Size of the heat transfer layer (mm)
25
In this paper, reflooding experiments were performed on a uniformly heated bank of cylinders. However, since all RTDs can be
controlled individually, this setup can be used to study the influence of heterogeneous heating and the impact of a timedependent heating power Q, situations that one may encounter
in practice, for instance in the context of nuclear reactor severe
accidents. This will be the subject of future studies. Furthermore,
image processing techniques are currently being developed to estimate the local void fraction field from the high speed visualizations, which would provide the necessary data for subsequent
quantitative modeling analysis.
vl = 1.25 mm/s
vl = 3.36 mm/s
vl = 7.58 mm/s
vl = 11.8 mm/s
vl = 16.0 mm/s
vl = 20.2 mm/s
20
15
10
Acknowledgments
5
0
0.1
1
10
100
Time (s)
Fig. 10. Vertical extent, Lhtl , of the heat transfer layer as a function of time, for
different fluid injection velocities v l .
4. Conclusion and perspectives
The experimental setup described in this article has been developed to study reflooding with internal boiling of a heating model
porous medium with mm-scale pores. The main objective is to provide precise data both at the pore and macroscopic scales, with a
perspective to better understand two-phase flows in porous media
with intense liquid–vapor phase change. The main element of the
apparatus is a slightly disordered bundle of cylinders placed
between two plates, thus forming a nearly two-dimensional particle bed. Each constitutive element is a cylindrical RTD, used both as
a temperature probe and a heating element. An interactive electronic device allows real-time control of the heating power transferred to each cylinder.
Temperature evolution of the heating cylinders, together with
direct visualization, enabled to identify three distinct macroscopic
regions and the heat transfer processes involved. First, in the
superheated vapor-dominated zone with dispersed small liquid
drops, the solid matrix is cooled down by vapor forced convection,
conduction through the walls and sputtering of fluid droplets from
the quenching front. Second, upstream of the quenching front, the
major amount of heat is extracted in a so-called heat transfer layer.
In this intense boiling region, the strong vapor bubbles confinement furthers coalescence, leading to the formation of vapor slugs,
which is reminiscent of microscopic boiling phenomenology.
Finally, in the thermally established reflooded zone, classical convective nucleate boiling with isolated nucleation sites is observed.
Thermal measurements also enabled to measure the quenching
and cooling fronts velocities for different heating powers and injection flow rates. These results from thermal measurements and
direct visualizations are similar to those inferred from thermocouple data in 3D particle beds, as far as the temperature evolution at a
given location within the bed is concerned or, e.g., regarding the
quenching front velocity evolution with injection velocity
[37,14,15]. That reinforces the proposed idea that the observations
made on the macroscopic behavior and the reflooding dynamics in
this model porous medium are representative of the physics of
two-phase flows in 3D debris beds with internal boiling. The
experimental setup, as well as the already obtained results, may
therefore be very useful to develop more accurate pore-scale or
macro-scale models and hence improve nuclear safety analysis or
any other application in which intense boiling in porous media
plays a major role.
Authors are particularly grateful to Ruddy Soeparno who contributed to the technical design and realization of the experimental
setup and to Hervé Ayroles who built the electronic modules controlling the RTDs. Thanks also go to Electricité de France (EDF) for
funding this research project.
References
[1] V. Dhir, I. Catton, Dryout heat fluxes for inductively heated particulate beds, J.
Heat Transfer 99 (2) (1977) 250–256.
[2] H. Bau, K.E. Torrance, Boiling in low-permeability porous materials, Int. J. Heat
Mass Transfer 25 (1982) 45–55.
[3] K. Udell, Heat transfer in porous media heated from above with evaporation,
condensation, and capillary effects, J. Heat Transfer 105 (3) (1983) 485–492.
[4] A. Reed, E. Bergeron, K. Boldt, T. Schmidt, Coolability of UO 2 debris beds in
pressurized water pools: DCC-1 and DCC-2 experiment results, Nucl. Eng. Des.
97 (1) (1986) 81–88.
[5] A. Stubos, J.-M. Buchlin, Vapour channels in boiling, unconstricted particle
beds – effect on the dryout heat flux, Int. J. Multiphase Flow 20 (1) (1994) 131–
152.
[6] L. Barleon, H. Werle, Dependence of Dryout Heat Flux on Particle Diameter for
Volume-and Bottom-Heated Debris Beds, Kernforschungszentrum Karlesruhe,
1981.
[7] V. Tung, V. Dhir, D. Squarer, Quenching by top flooding of a heat generating
particulate bed with gas injection at the bottom, in: The Sixth Meeting on
Debris Coolability, November 7–9, 1984, Los Angeles, California, Report EPRI
NP-4455, Palo Alto, California, March 1986.
[8] T. Ginsberg, J. Klein, J. Klages, Y. Sanborn, C. Schwarz, J. Chen, L. Wei,
Experimental and analytical investigation of quenching of superheated debris
beds under top-reflood conditions. Final report, Tech. rep., Brookhaven
National Lab., Upton, NY (USA) (1986).
[9] K. Hu, T. Theofanous, On the measurement and mechanism of dryout in
volumetrically heated coarse particle beds, Int. J. Multiphase Flow 17 (4)
(1991) 519–532.
[10] N. Tutu, T. Ginsberg, J. Klein, J. Klages, C. Schwarz, Debris Bed Quenching Under
Bottom Flood Conditions, Brookhaven National Laboratory, Upton, New York,
1984. Technical Report.
[11] P. Schäfer, M. Groll, R. Kulenovic, Basic investigations on debris cooling, Nucl.
Eng. Des. 236 (19) (2006) 2104–2116.
[12] I. Lindholm, S. Holmström, J. Miettinen, V. Lestinen, J. Hyvärinen, P.
Pankakoski, H. Sjövall, Dryout heat flux experiments with deep
heterogeneous particle bed, Nucl. Eng. Des. 236 (19) (2006) 2060–2074.
[13] K. Atkhen, G. Berthoud, SILFIDE experiment: coolability in a volumetrically
heated debris bed, Nucl. Eng. Des. 236 (19–21) (2006) 2126–2134.
[14] A. Bachrata, F. Fichot, G. Repetto, M. Quintard, J. Fleurot, Quench front
progression in a superheated porous medium: experimental analysis and
model development, J. Energy Power Eng. 7 (2013) 514–523.
[15] N. Chikhi, T. Garcin, F. Foubert, P. March, F. Fichot, First experimental results of
large scale debris bed reflood tests in the pearl facility, in: 16th International
Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH 16), 2014.
[16] A. Stubos, J.-M. Buchlin, Analysis and numerical simulation of the
thermohydraulic behaviour of a heat dissipating debris bed during power
transients, Int. J. Heat Mass Transfer 36 (5) (1993) 1391–1401.
[17] R. Lipinski, A coolability model for postaccident nuclear reactor debris, Nucl.
Technol. (United States) 65 (1) (1984) 53–66.
[18] T. Schulenberg, U. Müller, An improved model for two-phase flow through
beds of coarse particles, Int. J. Multiphase Flow 13 (1) (1987) 87–97.
[19] V. Tung, V. Dhir, A hydrodynamic model for two-phase flow through porous
media, Int. J. Multiphase Flow 14 (1) (1988) 47–65.
[20] K. Atkhen, G. Berthoud, Experimental and numerical investigations on debris
bed coolability in a multidimensional and homogeneous configuration with
volumetric heat source, Nucl. Technol. 142 (3) (2003) 270–282.
[21] M. Bürger, M. Buck, W. Schmidt, W. Widmann, Validation and application of
the WABE code: investigations of constitutive laws and 2D effects on debris
coolability, Nucl. Eng. Des. 236 (19) (2006) 2164–2188.
[22] F. Fichot, F. Duval, N. Trégourès, C. Béchaud, M. Quintard, The impact of
thermal non-equilibrium and large-scale 2D/3D effects on debris bed
reflooding and coolability, Nucl. Eng. Des. 236 (19–21) (2006) 2144–2163.
[23] F. Duval, F. Fichot, M. Quintard, A local thermal non-equilibrium model for
two-phase flows with phase-change in porous media, Int. J. Heat Mass Transfer
47 (3) (2004) 613–639.
[24] S. Whitaker, The Method of Volume Averaging, vol. 13, Springer, Netherlands,
1999.
[25] S. Whitaker, Flow in porous media II: the governing equations for immiscible,
two-phase flow, Transp. Porous Media 1 (2) (1986) 105–125.
[26] Y. Davit, B. Wood, G. Debenest, M. Quintard, Correspondence between oneand two-equation models for solute transport in two-region heterogeneous
porous media, Transp. Porous Media 95 (2012) 213–238, http://dx.doi.org/
10.1007/s11242-012-0040-y.
[27] Y. Davit, M. Quintard, Handbook of Porous Media, third ed., Theoretical
analysis of transport in porous media: Multi-Equation and Hybrid Models for a
Generic Transport Problem with Non-Linear Source Terms, Taylor & Francis,
2015. ch. 7.
[28] S. Liter, M. Kaviany, Pool-boiling CHF enhancement by modulated porouslayer coating: theory and experiment, Int. J. Heat Mass Transfer 44 (22) (2001)
4287–4311.
[29] V. Dhir, Boiling heat transfer, Annu. Rev. Fluid Mech. 30 (1) (1998) 365–401.
[30] V. Dhir, Mechanistic prediction of nucleate boiling heat transfer-achievable or
a hopeless task?, J Heat Transfer 128 (1) (2006) 1–12.
[31] P. Kew, K. Cornwell, Correlations for the prediction of boiling heat transfer in
small-diameter channels, Appl. Therm. Eng. 17 (8–10) (1997) 705–715.
[32] J. Thome, Boiling in microchannels: a review of experiment and theory, Int. J.
Heat Fluid Flow 25 (2) (2004) 128–139.
[33] P. Cheng, H. Wu, Mesoscale and Microscale Phase-Change Heat Transfer,
Advances in Heat Transfer, vol. 39, Elsevier, 2006, pp. 461–563.
[34] T. Ginsberg, J. Klein, C. Schwarz, J. Klages, Transient core debris bed heat
removal experiments and analysis, in: International Meeting on Thermal
Nuclear Reactor Safety, 1982.
[35] M. Rashid, R. Kulenovic, E. Laurien, Experimental results on the coolability of a
debris bed with down comer configurations, Nucl. Eng. Des. 249 (2012) 104–
110.
[36] C. Wang, V. Dhir, An experimental investigation of multidimensional
quenching of a simulated core debris bed, Nucl. Eng. Des. 110 (1) (1988) 61–
72.
[37] G. Repetto, T. Garcin, S. Eymery, F. Fichot, Experimental program on debris
reflooding (PEARL) – results on PRELUDE facility, Nucl. Eng. Des. 264 (2013)
176–186.
[38] F. Fichot, A. Bachrata, G. Repetto, J. Fleurot, M. Quintard, Quenching of a highly
superheated porous medium by injection of water, Journal of Physics:
Conference Series, vol. 395, IOP Publishing, 2012, p. 012144.
[39] R. Roser, B. Thonon, P. Mercier, Experimental investigations on boiling of npentane across a horizontal tube bundle: two-phase flow and heat transfer
characteristics, Int. J. Refrig. 22 (7) (1999) 536–547.
[40] L. Liao, Z.-H. Liu, Enhanced boiling heat transfer of the compact staggered tube
bundles under sub-atmospheric pressures, Heat Transfer Eng. 28 (5) (2007)
444–450.
[41] L. Aprin, P. Mercier, L. Tadrist, Local heat transfer analysis for boiling of
hydrocarbons in complex geometries: a new approach for heat transfer
prediction in staggered tube bundle, Int. J. Heat Mass Transfer 54 (19) (2011)
4203–4219.
[42] J. Broughton, P. Kuan, D. Petti, E. Tolman, A scenario of the three mile island
unit 2 accident, Nucl. Technol. 87 (1989) 34–53.
[43] R. McCardell, M. Russell, D. Akers, C. Olsen, Summary of TMI-2 core sample
examinations, Nucl. Eng. Des. 118 (3) (1990) 441–449.
[44] E. Coryell, D. Akers, C. Allison, M. Carbonneau, R. Hobbins, J. Hohorst, S. Jensen,
S. Modro, Summary of important results and SCDAP/RELAP5 analysis for OECD
LOFT experiment LP-FP-2, Technical Report NUREG/CR-6160, Washington D.C.,
1994.
[45] R. Manzel, C. Walker, EPMA and SEM of fuel samples from PWR rods with an
average burn-up of around 100 MWd/kgHM, J. Nucl. Mater. 301 (2) (2002)
170–182.
Download